Comment by charles-salvia | Hacker News Reader

Comment by charles-salvia | Hacker News Reader

If you trained a neural network to recognize "high powered tech executives", it will likely start picking up on features associated with 30 to 40 year old (mostly white) males wearing sweaters or "business casual" attire. It probably would report a low value if you showed it a minority female. (Google "tech executive" on Google images to test this out.)

Human beings are really advanced neural networks, trained on years of sensory data, along with lots of bias factors. This is why things like sexism and racism are often self-perpetuating. People aren't used to commonly seeing a female, minority woman in a high position, so they don't expect it, and therefore that expectation becomes part of their bias in the future.

There are only two solutions:

(1) Do things to create more minority female tech executives, so we all expect it.

(2) Do things to make people better at hiding their biases.

Obviously, we need to do both if this situation is ever going to change.

Couldn't agree more. The solution isn't to make this assistant apologize profusely. It's a reasonable mistake to make. Women of color are extraordinarily unusual in this position.

This isn't an instance of bias. It's a mistake based on simple base rate prevalence. No harm was done here and no harm was meant. Now, that doesn't mean that all forms of what people have taken to calling 'microaggressions' are harmless - some of them are harmful. This is just pretty clearly not one of them.

Option 3 is to re-train. Every time we see a woman of color think to ourselves "She's probably a powerful executive" and every time we see a white man in a suit we think "he's probably an executive assistant." Like a visualization technique.

The reality is that we're all biased but we're also malleable. If we understand how our biases manifest, we also understand that biases are pretty resilient to reason and goals. They need to be re-trained or manipulated (like through priming and other bias inducing mechanisms.)

This is what bothers me about unconscious bias training. They show that everyone is biased. But the solutions fail to acknowledge that the system can't change just through positive intention.

Standing out of the crowd has benefits and drawbacks.

You might not expect the black female to be a "high powered tech executive", but on the other hand, when you see the whole list of "high powered tech executives", you will probably remember the black female better than any random 40 year old white male.

I think treating (1) by lowering requirements is bad because now people will stereotype that that person got there on ez mode.

(2) is bad because people will inevitably make accidental genuine mistakes and it perpetuates the whole hostile PC culture where people are afraid of talking about anything and everything. If there is no social punishment then enforcing is impossible.

I think the solution is in changing culture. I.E.: make it like China or former USSR. People got jobs in in-demand fields, not in what they were interested in.

Most qualified people rose to the top primarily based on aptitude alone. That way there would be no need for thought police.

I think the solution is in changing culture. I.E.: make it like China or former USSR. People got jobs in in-demand fields, not in what they were interested in.

That sounds like a totalitarian state that nobody wants to live in. That might get rid of the thought police but you then you have job police, yeah?

The whole point is changing it in such a way that everyone "wants" to do these in-demand jobs. Chinese students aren't unhappy with their job choices.

Edit:

I think the easiest way is to simply not force people to self-answer an abstract question of "What are you interested in?" First of all, it's almost impossible to answer it correctly because there is no way you can sample all possible career paths. Secondly, if they don't ask themselves this question, they'll find interesting things in the general job area they're given.

We could similarly restrict all food production to rice and people would get used to it and even enjoy it, but that doesn't mean it is a solution towards a more diverse diet.

You are fundamentally talking about increasing diversity in one area by reducing choice in another. Isn't that antithetical?

edit: your edit makes more sense :)

Thanks!

Regarding food vs jobs, I think it's not a very good analogy because

a) almost every job has a lot to it that could be interesting for anyone

We are not limiting all jobs to "Wordpress site template maintainer for small e-commerce sites" or "MySQL DB engineer specializing in scaling and indexing databases with mostly JSON-based tables". If you are a front-end web dev, you could choose to end up specializing in front-end driven analytics or SVG drawing or something higher-level like d3 plotting or anything in HTML5 games or photoshop-to-code conversion or migrate to photoshop-based design etc etc

B) The goal is not to maximize diversity of food or diversity of jobs. I am not implying avoiding increasing specialization.

Tall white males with strong jaws and a full head of hair, preferably. Supermen, if you will (or Batmen, if you prefer).

The people who run tech companies are mostly similar demographics. They make a lot of short-sighted decisions.

Make your leadership pool more diverse in gender, race, and (more importantly) life experience and things will change. An artificial solution is only going to be top down, engineered by the very execs who have limited experience.

The bigger problem is the self-perpetuating vortex that is Silicon Valley. Mix it up.

While micro-aggressions suck, I agree with your sentiment. This is the problem with PC, it is treating the symptoms without addressing the fundamental causes.

Reply on news.ycombinator.com